Precision GPS experiment #1

This article reports an experiment to evaluate the usefulness of precision GPS for the purpose of location data for automated antenna field strength surveys.

The experiment was conducted with the rover located in a fixed location 13km North of the reference station at Symonston and with very wide view of the sky, about 7:00 am 07/04/2016.

Only GPS satellites were used for the rover.

The software was RTKLIB v2.4.3b8.

IMG_0505

The GPS was a UBLOX LEA-6T with a small patch antenna (as sold for small UAVs). The LEA-6T provides binary data as used by RTK for carrier phase measurements. Above is the GPS and a USB-RS232 adapter.

Single mode (or uncorrected)

The first experiment was to observe the scatter of position reports using single or uncorrected mode of RTKLIB.

Screenshot - 07_04_16 , 06_55_50

Above, a screen shot of RTNAVI in single mode (no reference station input).

Screenshot - 07_04_16 , 06_55_50_ver001

Above, a screen shot of RTPLOT in single mode (no reference station input) over about 10 minutes. The rapid variation in position reports limits the application. The RMS error was not calculated, but it can be seen to be several metres and would be more with more restricted sky access or with poorer S/N.

Screenshot - 07_04_16 , 06_52_12

Kinematic mode

Above, a screen shot of RTNAVI in kinematic mode (uses reference station input).

Screenshot - 07_04_16 , 06_52_12_ver001

Above, a screen shot of RTPLOT in Kinematic mode (using reference station input) over about 10 minutes.

The variation in position reports is much slower than the single case and again limits the application. The RMS error was not calculated, but it can be seen to be perhaps a couple of metres, a little better than single mode, and would be more with more restricted sky access or with poorer S/N.

We might also expect increased RMS error with a moving rover

Static mode

A short test was conducted in Static mode (using the reference station and depending on a stationary rover), and RMS error was a little better than Kinematic mode, but Static mode has no application in a moving field strength observation station.

Distance to reference station

Access to nearby reference stations can be an issue. In this experiment, the observation station was located near a reference station but that may not be generally convenient and increased error accrues from distance to error station.

Baseline distance contributes to error, and it seems a figure of 1ppm or 1mm/km is commonly used (Henning 2011). At my home location 150km from the Symonston reference station, baseline error contribution would be around 150mm.

Options include establishing a local reference station.

Conclusions

Single mode is simple, has rapid variation and an RMS error of the order of 5m with excellent sky view, poorer with restricted view or poorer S/N.

Kinematic mode provides better accuracy but with a much more complicated equipment setup.

The need to establish a local reference station in my own situation is a further complication.

Overall I relegate the use of GPS locations for antenna surveys that can tolerate location error of 10m, and kinematic GPS techniques provide only a very small improvement at the cost of much greater complexity.

Acknowledgements

Thanks go to Geoscience Australia for free access to an NTRIP feed for this research.

References

  • Henning, W. Aug 2011. User guidelines for single base real time GNSS positioning. NASA.